分形约束二维Otsu红外图像分割方法

Infrared Image Segmentation Based on 2D Otsu with Fractal Constraining

提出了一种分形约束Otsu阈值分割算法。该方法结合图像的灰度分布和像素间空间分形纹理信息,在二维类间方差准则取最大值时得到一个二维分割阈值,从而实现红外图像的自适应分割。将该方法应用到低对比度、低信噪比、边缘模糊的红外图像分割中,并与传统的二维Otsu分割方法作了比较。结果表明,该方法在分割效果和抗噪能力等方面均得到了明显的改善。

An improved two-dimensional Otsu segmentation algorithm for IR images,which is constrained by fractal,is proposed.This method combines the grayscale distribution of the image,and the fractal texture information between pixels.It computes the two-dimensional threshold by maximizing tthe inter-class variance to achieve the adaptive threshold segementation of the infrared image.This algorithm is applied to low contrast,low signal noise ratio,blurred infrared image segmentation,and the result shows that,comparing with the traditional two-dimensional Otsu segmentation methods,our algorithm have better segmentation result and noise resistance ability.